How does Level 3 autonomous driving work in EVs?

08 Ağu 02:33

Same here, I also experienced a Level 3 test drive in an EV prototype; the feeling of the car taking control without touching the wheel is impressive, but it only works under very specific conditions. Regarding hardware, Level 3 vehicles typically combine short- and long-range LIDAR, multiple 77 GHz radars, and at least three high-resolution cameras covering the front, sides, and rear. All of this feeds into a central processing unit (high-performance GPU/CPU) that needs a constant power source—in EVs, this is managed with a dedicated 12V battery and redundant power architecture so that if the main system fails, control is immediately returned to the driver. Additionally, HD map calibration and 5G connectivity enable real-time road data updates. On the software side, it consists of sensor fusion layers, object detection, path planning, and a driver monitoring module. The latter tracks attention via cabin cameras and seat sensors; when the car detects the driver is ready, it issues visual and auditory alerts, and if there’s no response, it slows down and hands back control through progressive torque assistance. The entire stack is certified under functional safety standards (ISO 26262) and updated via OTA to incorporate AI improvements and new features. Current limitations are still quite clear: adverse weather (fog, heavy snow), poor signage, and unmapped roads can disable Level 3. Regulatory restrictions and the requirement for the driver to keep their eyes on the road also reduce the "true autonomy." Looking ahead, the integration of low-cost LIDAR, more robust deep learning AI, and V2X (vehicle-to-infrastructure) communication are expected to expand operational domains to more complex highways and eventually smoother transitions to Level 4. In our workshop, we’re already testing perception modules that promise to cut detection latency by 30%, which could be a game-changer for making Level 3 a standard feature in next-gen EVs.